對於馬賽克影像的錯誤修正演算法

Abstract

大部分的數位相機使用單CCD/CMOS感應器來得到彩色馬賽克影像，其中每個點只有一個顏色，當傳輸會產生錯誤和雜訊，本論文提出了一個新的對於馬賽克影像有效的錯誤修正演算法，為了同時移除頻譜域及空間域上的統計性多餘量，我們使用了四階的2DMallat小波轉換來對馬賽克資料做去相關的動作，在各個分開的子頻中，我們使用資料隱藏的技術以用來修復錯誤，就我們所知這是第一次有人提出在馬賽克影像上的錯誤修復方法，而在一些測試的圖片中，實驗結果顯示我們的方法在視覺效果和圖片品質都優於內插法和平均法。

Most digital cameras use a single CCD/CMOS sensor array to capture color mosaic images, where each pixel has only one measured color. Considering the transmission error and the noise influence during the communication, this thesis presents a new and efficient error concealment method for mosaic images. For removing statistical redundancies in both spectral and spatial domains, we perform a four–level 2–D Mallat packet wavelet transform to decorrelate the mosaic data. Then, based on the partitioned subbands, an efficient data embedding technique is presented to conceal errors. To the best of our knowledge, this is the first time that such an error concealment method is presented for mosaic images. Under some test mosaic images, experimental results demonstrate that both the image quality and the visual effect of our proposed error concealment method are superior to the Inter method and the Mean method.